| Concrete bridge decks require periodic condition assessment and preventive maintenance to extend their useful lifespan. Nondestructive evaluation methods such as Ground Penetrating Radar (GPR) are slowly beginning to replace or complement the manual (visual) assessment of bridge conditions for detecting defects at their early stages. However, GPR scans of bridge decks are frequently cluttered with high amplitude reflections from known parts of the bridge deck, which make the detection of defects' low amplitude reflections difficult. One such known part is the embedded steel reinforcement bars known as rebar.;This dissertation presents a novel approach to the automated detection of defects in concrete bridge decks by removing known reflections such as rebar from GPR scans of reinforced concrete bridge decks. The algorithm detects reflections from rebar with a frequency-domain pulse detection method, groups detected pulses into clusters, interpolates synthetic rebar reflections based on each cluster, and subtracts the synthetic rebar reflection from the original GPR scan data. This algorithm will facilitate the automated, non-destructive condition assessment of bridge decks. |
